The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large volumes of untrusted novel text as its primary function. This creates a surface for indirect prompt injection if a novel contains adversarial instructions. However, the skill implements a structured pipeline using CSV, JSON, and specific Markdown schemas, and its processing scripts do not interpret the novel content as executable code or control instructions.
[COMMAND_EXECUTION]: The script scripts/render_relation_chart.py contains a call to subprocess.run to execute the fc-list command. Technical review confirms this is a safe, static invocation used solely for detecting system fonts to ensure proper rendering of Chinese characters in relationship charts, and it does not pass user-supplied arguments to the shell.
[SAFE]: The skill operates entirely within the local environment, reading novel files and writing analysis results to the project directory. No network operations, credential access, or persistence mechanisms were detected. The use of local Python scripts for data processing is consistent with the skill's stated purpose as a literary analysis tool.

story-long-analyze